9 Replies Latest reply on Feb 3, 2015 2:01 PM by Jared Conway

    trouble with flow simulation

    Dan Hofstetter

      I have been having some trouble with a flow simulation where I am getting a very high velocity near an outlet volume flow boundary condition.  I have tried many things to eliminate this:  using overlapping lids at a flow opening and defining outlet volume flow, velocity, and mass flow rates, as well as leaving a solid boundary and defining a split line to use as the outlet.  I also noticed when inspecting average face velocities at the outlet that the Area (Fluid) matches the geometry correctly, i.e. an 8" diameter circle is shown as 0.349 sq.ft., but the surface area shown next to the velocity is always lower (0.346 sq.ft.).  To work around this, I added a local initial mesh to the outlet face and set the All Cells refinement level to 4 or 5.  This helps to capture the surface area, but I'm still getting some weird numbers.

       

      For instance, with a 14.5" diameter flow outlet at 1,302 cfm, the velocity should equal 18.9 ft/s, but I'm seeing min = 18.97, max = 87.4, average = 27.2 ft/s.  I think that one high velocity is throwing it all off, but I can't seem to get rid of it.  When I look at the velocity cut plot and show the min/max locations, I can see that this velocity occurs in one cell that is near the edge of the outlet.  Have any of you seen this problem, and can you recommend a fix?  I've attached my 2015 version flow model in case anyone wants to take a look.

       

      Thanks,

       

      Dan

        • Re: trouble with flow simulation
          Jared Conway

          improving the number of cells in that region doesn't help solve the issue with a couple of cells being at very high velocity? are the majority of the cells the right velocity?

           

          why are you using an outlet volume flow rate? this is generally not recommended by the developers because of the assumptions it makes regarding the velocity distribution.

           

          i haven't looked in the model but it looks like your have a porous medium. have you already simplified your model down to just the bare bones to see if the problem still exists. ie just an inlet and outlet, flow only (no heat) problem?

            • Re: trouble with flow simulation
              Jared Conway

              also, there are about 11 studies in this file, which one should we be looking at?

                • Re: trouble with flow simulation
                  Jared Conway

                  first quick look at the 7th study

                   

                  i would highly recommend getting rid of all the overlapping components, especially between the porous and solids.

                    • Re: trouble with flow simulation
                      Jared Conway

                      and also add some good goals at the inlet and outlet to make sure that they are converging

                        • Re: trouble with flow simulation
                          Jared Conway

                          ran the model for a little while, i see the same issue. i think you have a mesh related issue here. remember that flow simulation likes cells that are almost square. the ones in that area are not very pretty.

                           

                          The other thing you may want to consider is extending that outlet to move the boundary condition away from the area of interest. if there is a computational issue, at least it won't affect where you are trying to measure things.

                           

                          but i would also go through the other troubleshooting steps of minimizing the elements to see if it helps. i have a feeling that outlet volume flow rate and an inlet volume flow rate and also a porous medium might be a challenging combination. each area are almost like separate fluid domains that need to be calculated separately.

                            • Re: trouble with flow simulation
                              Dan Hofstetter

                              Thanks for taking a look!  I cut through the porous medium so there were no overlapping solid bodies, but no improvement.  I set all temperatures = 68F.  Is there an option to turn off heat altogether?  I've never found one for isothermal simulations.  I also have not found a way to turn off friction other than defining ideal walls everywhere.  I removed all gases except for air, so the concentration equation should be dropped (if that's how Flow Simulation works).  And I simplified the model as much as possible everywhere else.

                               

                              I set up another similar, simplified problem, and get the same high velocity close to the fan face.  That high velocity is always in another direction, i.e. not parallel to the flow from the boundary condition.  Am I correct in thinking that if I get even one crazy cell value that the calculation is probably wrong?  It seems like the high cell makes convergence take much longer, too.

                               

                              I might try to set up a unidirectional porous medium attached to that face to prevent flow from the other directions, but that might just chase the problem further from the boundary condition.

                            • Re: trouble with flow simulation
                              Dan Hofstetter

                              What inlet/outlet goals would you suggest?  I've used one to look at {Outlet Mass Flow Rate}-{Inlet Mass Flow Rate} before to check for continuity of mass, but it doesn't seem to be necessary since those values are nearly equal in these simulations.

                        • Re: trouble with flow simulation
                          Dan Hofstetter

                          Increasing the number of cells in that region seems to make the problem worse.  The few times I have been able to run a simulation where the velocities were at expected values were when I had no refinement in the mesh in that area.  The majority of the cells are at the right value, but that one high cell pulls the average velocity up too much, and I'm afraid it might have a negative impact on the results.

                           

                          The reason I'm using velocity there is that is what was measured.  I have measured inlet and outlet velocities, so it seems appropriate to use those to define the boundary conditions.  I did try using mass flow rate and volume flow rate, but they give me the same trouble.